#Mendelian Randomization Analysis: HDL Cholesterol and Alzheimer's Disease By Dr. Shea Andrews, generated on 2018-09-08
## [1] FALSE
HDL Cholesterol Willer et al Nat Genet 2013: To identify new loci and refine known loci influencing these lipids, we examined 188,577 individuals using genome-wide and custom genotyping arrays. We identify and annotate 157 loci associated with lipid levels at P < 5 × 10−8, including 62 loci not previously associated with lipid levels in humans.
Late Onset Alzheimer’s disease Lambert et al 2013: The International Genomics of Alzheimer’s Project (IGAP) is a meta-analysis of 4 previously published GWAS datasets: the European Alzheimer’s Disease Imitative (EADI), the Alzheimer Disease Genetics Consortium (ADGC), Cohorts for Heart and Aging Research in Genomic Epidemiology (CHARGE), and Genetic and Environmental Risk in AD (GERAD) and includes a sample of 17,008 LOAD cases and 37,154 cognitively normal elder controls. Participants in IGAP were of European ancestry, the average age was 71 and 58.4% of participants were women.
Alzheimer’s Age of Onset Surivial Huang et al 2017: A GWAS of age of onset in LOAD was conducted in 14,406 AD case samples and 25,849 control samples from the IGAP using Cox proportional hazard regressions. Participants were of European ancestry, in cases the the average AAO was 74.8 and 61.7% were women, in controls the average AAE was 79.0 and 59.6% were women.
CSF Ab42, tau & ptau Deming et al 2017: A GWAS of CSF AB42, ptau and tau levels (pg/mL) was conducted in 3,146 participants. Participants were of Eurpean ancestry.
Hippocampal Volume Hibar et al 2017: A GWAS of hippocampal volume perfomed in 26,814 (ENIGMA and CHARGE consortiums) individules of European Ancestry discovered 9 independent loci.
LD Clumping: For standard two sample MR it is important to ensure that the instruments for the exposure are independent. LD clumping can be performed using the data_clump function from the TwoSampleMR package, which uses EUR samples from the 1000 genomes project to estimate LD between SNPs and amonst SNPs that have and LD above a given threshold, only the SNP with the lowest p-value will be retained.
Proxy SNPs: SNPs associated with HDL Cholesterol were extracted from the GWAS of LOAD, AAOS, AB42, ptau and tau. Where SNPs were not available in the outcome GWAS, the EUR thousand genomes was queried to identified potential proxy SNPs that are in linkage disequilibrium (r2 > 0.8) of the missing SNP.
HDL Cholesterol: 5412 SNPs (Table 1) were assoicated with were associated with HDL Cholesterol at p < 5e-6. After LD clumping, 5045 of 5412 SNPs were removed.
Of the the 367 SNPs associated with HDL Cholesterol, 347 were available in the LOAD GWAS (Table 2).
Table 2: SNPS associated with HDL Cholesterol avalible in LOAD GWASOf the the 367 SNPs associated with HDL Cholesterol, 364 were available in the AAOS GWAS (Table 3).
Table 3: SNPS associated withOf the the 367 SNPs associated with HDL Cholesterol, 291 were available in the CSF AB42 GWAS (Table 4).
Table 4: SNPS associated with HDL Cholesterol avalible in ab42 GWASOf the the 367 SNPs associated with HDL Cholesterol, 325 were available in the CSF Ptau GWAS (Table 5).
Table 5: SNPS associated with HDL Cholesterol avalible in Ptau GWASOf the the 367 SNPs associated with HDL Cholesterol, 291 were available in the CSF Tau GWAS (Table 6).
Table 6: SNPS associated with HDL Cholesterol avalible in tau GWASOf the the 367 SNPs associated with HDL Cholesterol, 348 were available in the Hippocampal volume (Table 7).
Table 7: SNPS associated with HDL Cholesterol avalible in Hippocampal volume GWASHarmonize the exposure and outcome datasets so that the effect of a SNP on the exposure and the effect of that SNP on the outcome correspond to the same allele. The harmonise_data function from the TwoSampleMR package can be used to perform the harmonization step, by default it try’s to infer the forward strand alleles using allele frequency information. EAF were not availbe in the IGAP summary statisitics, as such the allele frequencies reported in the AAOS anaylsis were used.
Pleiotropy was assesed using Mendelian Randomization Pleiotropy RESidual Sum and Outlier (MR-PRESSO), that allows for the evlation of evaluation of horizontal pleiotropy in a standared MR model. MR-PRESSO performs a global test for detection of horizontal pleiotropy and correction of pleiotropy via outlier removal.
HDL Cholesterol ~ LOAD: The MR-PRESSO global test for pleiotropy was significant (p = <4e-05). The following SNPs were removed due to pleiotropy: rs11604680, rs157580, rs17788930, rs2075650, rs4803760, rs4803763, rs5167, rs7412, rs76366838
HDL Cholesterol ~ AAOS: The MR-PRESSO global test for pleiotropy was significant (p = <4e-05). The following SNPs were removed due to pleiotropy: rs157580, rs2075650, rs4803763, rs5167, rs76366838
HDL Cholesterol ~ AB42: The MR-PRESSO global test for pleiotropy was significant (p = <4e-05).. The following SNPs were removed due to pleiotropy: rs157580, rs2075650
HDL Cholesterol ~ Ptau: The MR-PRESSO global test for pleiotropy was significant (p = <4e-05).. The following SNPs were removed due to pleiotropy: rs157580, rs2075650, rs7412
HDL Cholesterol ~ Tau: The MR-PRESSO global test for pleiotropy was significant (p = <4e-05).. The following SNPs were removed due to pleiotropy: rs1645890, rs2148489, rs7601692
HDL Cholesterol ~ hippocampal volume: The MR-PRESSO global test for pleiotropy was significant (p = 0.00168). The following SNPs were removed due to pleiotropy: rs2075650
To obtain an overall estimate of causal effect, the SNP-exposure (Major Depressive Disorder) and SNP-outcome coefficients (Alzheimer’s disease and Alzheimer’s Age of Onset) were combined in 1) a random-effects meta-analysis using an inverse-variance weighted approach (IVW); 2) a Weighted Median approach; 3) and Egger Regression. IVW is equivalent to a weighted regression of the SNP-outcome coefficients on the SNP-exposure coefficients with the intercept constrained to zero. This method assumes that all variants are valid instrumental variables based on the Mendelian randomization assumptions. The causal estimate of the IVW analysis expresses the causal increase in the outcome (or log odds of the outcome for a binary outcome) per unit change in the exposure. Weighted median MR allows for 50% of the instrumental variables to be invalid. MR-Egger regression allows all the instrumental variables to be subject to direct effects (i.e. horizontal pleiotropy), with the intercept representing bias in the causal estimate due to pleiotropy and the slope representing the causal estimate.
Figure 1 illustrates the SNP-specific associations with HDL Cholesterol versus the association between each SNP and risk of LOAD.
Fig. 1: Scatterplot of SNP effects for the association of Trait and LOAD
Figure 2 and Table 1 shows the SNP-specific effects and overall IVW, weighted median and Egger regression causal estimates of genetically predicted HDL Cholesterol on risk of LOAD.
Fig. 2: Forrest plot of Wald ratios and 95% CIs for SNP-specific and overall IVW, Weighted median and Egger associations
Figure 3 shows a funnel plot to detect pleiotropy and Table 2 show the results of Cochrans Q heterogeneity test to assess for the presence of pleiotropy.
Fig. 3: Funnel plot of the Trait – LOAD causal estimates against their precession
Figure 4 illustrates the SNP-specific associations with HDL Cholesterol versus the association between each SNP and AAOS.
Fig. 4: Scatterplot of SNP effects for the association of Trait and AAOS
Figure 5 and Table 4 shows the SNP-specific effects and overall IVW, weighted median and Egger regression causal estimates of genetically predicted HDL Cholesterol on AAOS.
Fig. 5: Forrest plot of Wald ratios and 95% CIs for SNP-specific and overall IVW, Weighted median and Egger associations
Figure 6 shows a funnel plot to detect pleiotropy and Table 5 show the results of Cochrans Q heterogeneity test to assess for the presence of pleiotropy.
Fig. 6: Funnel plot of the traitohol Conumption – AAOS causal estimates against their precession
Figure 1 illustrates the SNP-specific associations with HDL Cholesterol versus the association between each SNP and AB42.
Fig. 1: Scatterplot of SNP effects for the association of trait and AB42
Figure 2 and Table 7 shows the SNP-specific effects and overall IVW, weighted median and Egger regression causal estimates of genetically predicted HDL Cholesterol CSF AB42 levels.
Fig. 2: Forrest plot of Wald ratios and 95% CIs for SNP-specific and overall IVW, Weighted median and Egger associations
Figure 3 shows a funnel plot to detect pleiotropy and Table 8 show the results of Cochrans Q heterogeneity test to assess for the presence of pleiotropy.
Fig. 3: Funnel plot of the trait – AB42 causal estimates against their precession
Figure 1 illustrates the SNP-specific associations with HDL Cholesterol versus the association between each SNP and risk of Ptau.
Fig. 1: Scatterplot of SNP effects for the association of trait and Ptau
Figure 2 and Table 9 shows the SNP-specific effects and overall IVW, weighted median and Egger regression causal estimates of genetically predicted HDL Cholesterol on risk of Ptau.
Fig. 2: Forrest plot of Wald ratios and 95% CIs for SNP-specific and overall IVW, Weighted median and Egger associations
Figure 3 shows a funnel plot to detect pleiotropy and Table 10 show the results of Cochrans Q heterogeneity test to assess for the presence of pleiotropy.
Fig. 3: Funnel plot of the trait – Ptau causal estimates against their precession
Figure 1 illustrates the SNP-specific associations with HDL Cholesterol versus the association between each SNP and CSF Tau levels.
Fig. 1: Scatterplot of SNP effects for the association of trait and Tau
Figure 2 and Table 12 shows the SNP-specific effects and overall IVW, weighted median and Egger regression causal estimates of genetically predicted HDL Cholesterol on risk of Tau.
Fig. 2: Forrest plot of Wald ratios and 95% CIs for SNP-specific and overall IVW, Weighted median and Egger associations
Figure 3 shows a funnel plot to detect pleiotropy and Table 13 show the results of Cochrans Q heterogeneity test to assess for the presence of pleiotropy.
Fig. 3: Funnel plot of the trait – Tau causal estimates against their precession
Figure 1 illustrates the SNP-specific associations with HDL Cholesterol versus the association between each SNP and hippocampal volume.
Fig. 1: Scatterplot of SNP effects for the association of trait and hippocampal volume
Figure 2 and Table 12 shows the SNP-specific effects and overall IVW, weighted median and Egger regression causal estimates of genetically predicted HDL Cholesterol on risk of hippocampal volume.
Fig. 2: Forrest plot of Wald ratios and 95% CIs for SNP-specific and overall IVW, Weighted median and Egger associations
Figure 3 shows a funnel plot to detect pleiotropy and Table 13 show the results of Cochrans Q heterogeneity test to assess for the presence of pleiotropy.
Fig. 3: Funnel plot of the trait – hippocampal volume causal estimates against their precession